1. Introduction
The electrical wiring interconnection system (EWIS) plays key roles in aircraft, which composes various cables and line devices to transmit electrical energy or signals in aircraft [
1]. Ensuring the stability of the EWIS is an important issue for the safety of aircraft. Statistics show that insulation wear accounts for 30% of the flaws in aircraft cables [
2,
3]. In addition, many accidents and aircraft failures are directly related to the flaws in the cable insulation of the EWIS, such as the American TWA B747 plane in 1996 and the Swissair MD-11 plane in 1998, both which had flight accidents due to flaws in the insulation layer of their aviation cables [
4]. Therefore, it is very important to investigate and develop the defect detection technology of the cable insulation layer to ensure the flight safety of civil aircraft.
Current aviation cable detection methods and equipment have relatively weak detection capabilities and poor detection results, which makes it difficult to ensure aircraft flight safety [
5,
6,
7]. Various nondestructive testing (NDT) methods for damaged cable detection have been reported. The time–domain reflection method of the reflectometer is sensitive for cable surface detection, but the requirement of cable surface cleanliness is high, and it is not suitable for in situ detection [
8,
9,
10]. Radiographic testing equipment is complex, with a certain radiation hazard [
11]. The capacitance detection method can only be used to measure the dielectric constant of a single cable insulation layer [
12]. Ultrasonic methods require a high cable surface cleanliness for the use of the couplant [
13,
14]. The eddy current method is not available for cases wherein the conductor inside the cable is broken, or for the detection of non-conductive insulation layers [
15]. Capacitive tomography is susceptible to complex environments in the detection of cable insulation defects [
16].
Most of the enterprises follow the traditional manual inspection methods, namely the visual inspection method and the megohmmeter or multimeter measurement methods [
17,
18]. The visual inspection method uses flashlights, reflectors, magnifying glasses, and other equipment to check the insulation layer of the cable section for cracks, wear, and other fault detection. The megohmmeter method mainly measures insulation resistance, and the multimeter method generally measures electrical parameters such as resistance, voltage, and current to determine whether each cable is on or off, one by one. The megohmmeter method and the multimeter methods have complicated procedures, low efficiency, and easy-to-miss detection. Some researchers have developed a hand-held aviation cable detector based on the time–domain reflectometry method [
19]. By transmitting a pulse signal to one end of the cable, the impedance mismatch characteristics of the fault point are used to obtain the reflected signal parameters of the fault point so as to determine the type of fault and the location of the fault point. This method is a typical single-ended test method which can effectively solve the problem of cable fault detection in “invisible and unreachable” locations on the aircraft. However, the reflectometer is very sensitive, so the cleaning of the cable surface and the connection between the reflectometer and the cable being tested should be very carefully prepared. In addition, this method is not suitable for the in situ detection of the cable during the service of the aircraft. White et al., of Johns Hopkins University in the United States, used three methods (infrared thermal imaging, time domain reflectometry, and pulsed radiography) to detect aviation and aerospace cables, and to detect short circuits in the cable bundles and insulation damage [
9]. Li et al. of Iowa State University used an lnductance, capacitance and resisitance (LCR) tester to measure the dielectric constants of polytetrafluorethylene (PTFE) and ethylene-tetrafluoroethylene (ETFE), which are insulating materials for aviation and aerospace cables. The dielectric constants reflect the performance degradation of the insulating materials under ambient temperature changes [
20]. Chen et al., from Iowa State University in the United States, designed and produced a capacitance probe for the quantitative non-destructive testing of the insulation performance of aviation cables, and judged whether the cable had insulation aging defects through the change of complex permittivity [
10].
Infrared detection technology is suitable for the rapid in situ detection of a large structure, and it plays an increasingly important role in fault detection in power systems, equipment, and cables [
17]. Sfarra et al. from Italy systematically investigated the infrared thermographic technique for defect detection in both numerical and experimental manners [
21,
22]. In this paper, infrared detection technology was used to detect the insulation layer of the EWIS cable. A halogen lamp was used to heat the cable. Temperature distribution on the surface of the cable was obtained by an infrared thermal imager, and then the integrity of the cable insulation layer was analyzed.
3. Experiment
3.1. Experimental System and Sample
The infrared nondestructive testing system is shown in
Figure 4. The external thermal excitation source (halogen lamp) heated the tested cable sample, and the heat flowed inside the cable, breaking the original thermal equilibrium state, and the infrared thermal imager obtained the cable. The sequence of surface heatmaps, through the analysis of the sequence of heatmaps, realized the identification of defects in the cable insulation layer. The halogen lamp power was 1000 W, which was set about 25 cm away from cable. The infrared thermal camera was a FLUKE TI200, which has a horizontal FOV of 45° (approximately), 160 pixels horizontally, and an IR lens of focal length 6.5 mm. The official IFOV from the specs sheet is 5.2 mrad, and a focus distance of 30 cm was used. The infrared spectral band was 7.5~14 μm. During the experiment, the infrared thermal imager was placed 30 cm away from the cable. Thermal image resolution, or spatial resolution, is an important parameter, e.g., field of view (FOV), instantaneous field of view (IFOV), and detector array, which were considered when choosing the infrared camera. These parameters can be used to indicate the ability of the camera to distinguish between two objects in the field of view, which primarily depends on the object-to-camera distance, lens system, and detector size [
24,
25,
26,
27]. The camera lens can be carefully adjusted to achieve a better image effect.
The cable sample model of the first group is MIL-W-22759/35-24, and the parameters are shown in
Table 2. The measured objects are the whole insulation cables with lengths of 45 cm and different diameters (Group #1 and Group #2). Artificial defects were made on the cables to simulate wear defects. The article defects were made in the surface of the cables. The specific values are shown in
Table 3. In order to measure the length of the defect, we fixed the five cables with a foam box and took the third cable as a reference cable.
The cable samples used in the second set of experiments were collected from a retired Boeing 747. The cables were routed separately. We checked the integrity of the tested samples. Before introducing the defects into the cables, we cleaned their surfaces and checked their integrity visually. The cable parameters are shown in
Table 4. Two defects were artificially created on the cables to simulate wear defects, and the lengths of the defects were measured by a ruler. The specific values are shown in
Table 5.
3.2. Initial Test
In this study, we recorded original heatmap for every time period of heating and then subtracted respective heatmaps from the thermograms with cables. We fixed the insulation cables with different defects inside an open box to ensure a consistent background of all the cables. Secondly, we processed the infrared images to minimize the interference of the complicated background temperature distribution. In this study, the temperature distribution on the tested samples with different heating times were firstly observed and analyzed for optimizing relative parameters to obtain the desired results.
The samples of Group #1 were heated for 80 s, and the distribution of the surface temperature of the cables when heated for 10 and 80 s was obtained by using an infrared thermal imager. The power of the halogen lamp was turned off to stop heating. An infrared thermal imager was used to obtain the distribution of the surface temperature of the cable when the cable was naturally cooled for 20 and 80 s. The values of time for heating and cooling were selected by referencing the manual introduction of the halogen lamp. In addition, we also compared the infrared images during the heating and cooling process. It was noticed that at these four different times, we could obtain the obvious differences of image features.
Figure 5 shows the temperature distribution of the cable surface at the four time points, respectively. After heating for 10 s, the temperature at the defect of the second cable and the fifth cable was significantly lower than the temperature at the defect-free place (
Figure 5a), indicating that the use of infrared detection technology could effectively detect the insulation layer of the cable defect. After heating for 80 s, the display of defects was more obvious (
Figure 5b). In the cooling stage, with the increase of cooling time, the temperature difference between the defect and the non-defect on the surface of the cable decreased.
It is important to note that the insulation cables of aircraft are generally placed in a limited space inside of the aircraft structure, without a relative ideal background. In addition, one of the objectives of this study was to study and remove the complicated background images for the improvement of infrared images of defects. Thus, we did not perform the testing in front of a cold wall with a high thermal mass.
The cable samples of Group #2 were heated and the temperature distribution of the cable surfaces was also obtained. The cable with the defect on the front was heated for 120 s, and the temperature distribution of the cable surface for the different heating times was obtained, as shown in
Figure 6. In the first 100 s heating process, both full-wear defects and half-wear defects were displayed. The full-wear defects were more obvious than the half-wear defects. When heated for 120 s, almost no defects were observed from the heatmap. We heated the cable with the defect on the reverse side for 110 s to obtain a heatmap sequence, as shown in
Figure 7. During the first 90 s heating process, full-wear defects and half-wear defects were also displayed, and after 110 s of heating, almost no defects could be observed from the heatmap. It is important to note that we intentionally did not set the non-focusing thermal irradiation to explore the influences on defect detection in this study, since in some practical testing, it cannot always ensure the thermal focusing. To improve the accuracy of defect identification, the infrared images were processed to reduce the interference of the background temperature.
Based on the comparisons, it was found that the defect could be clearly identified either in the front or back sides of the cables when we did not heat the cables for a long time period (in this study it was about 100 s). Thus, the obtained results suggest that defects existed in different positions on the cables could be identified with similar effects as in the case of the cables not being heated for a long time. It is important to note that we did not intentionally set the focusing condition to achieve the infrared images with a high quality. The current investigation is still working on the edge of focal distance for the significant effect of thermal focusing on infrared measurements.
4. Infrared Image Processing and Analysis
The infrared images obtained by the infrared thermal imager contain the interference of the background temperature with the complex distribution [
28]. In order to improve the accuracy of the defect identification, the infrared image was processed by commercial software. The linear interpolation method was used to obtain the background temperature data in the cable area from the background temperature data outside the cable area. We constructed the background temperature image and subtracted the original heat map from the background temperature image to obtain the differential image. The accuracy of the defect identification was improved by reducing the interference of the background temperature. Finally, the data of the cable area was extracted from the differential image to draw a curve to analyze the effect of the defects on the surface temperature of the cable. The infrared image processing flow is shown in
Figure 8.
The infrared images of the first group of cables heated for 80 s were taken as an example for image processing. The original heatmap is shown in
Figure 9a. The position of the edge of the cable area was extracted. The temperature data in the cable area was obtained from the temperature data outside the cable area through the linear interpolation method to construct a background temperature image, as shown in
Figure 9b. The original heat map was subtracted from the background temperature map to obtain a differential image without the background temperature, as shown in
Figure 9c. The results of image after processing show that the defects are more obvious in the image after removing the interference of the background temperature.
However, although the temperature change range at the cable defect in the heatmap is related to the length of the defect, the length of the defect cannot be effectively obtained by only observing the heat map and the differential image. Therefore, a signal processing program was used to extract the cable in the differential image (
Figure 9c). After obtaining
Figure 9c, we used commercial software (Matlab) to extract the image features, including the three sets of differential temperature data in the cable area in the differential images. Then, the three sets of differential temperature data at the center of the cable could be achieved by using those at the edge position of the cables. The averages of the three differential temperature data were calculated, and the differential temperature curve was plotted for analysis, as shown in
Figure 10. It was found that the length of the defect on the second cable and fifth cable was exactly equal to half of the number of pixels in the temperature anomaly range. The defect lengths of the second and fifth cables were 7.0 and 4.5 mm, respectively. Therefore, by counting the number of pixels in the temperature change area, the length of the defect could be obtained. When the insulation was worn but the conductor was not exposed (as in the first and fourth cables), no defect was observed in the heatmap. However, the temperature change at the defect could be observed in the differential temperature curve.
It should be noted that the numerical study provides us a qualitative suggestion about temperature distribution on the surface of the cables, which can guide the experiment setup. There are some differences between experiment and numerical model in terms of measurement setup or excitation time for the reason that it has an ideal background and heat dissipation from the cables to the air in the numerical simulation. However, in the experiment, the ideal background and uneven heat dissipation caused the obvious differences of the infrared detection results. Thus, we set the different heating times to obtain infrared images in the numerical and experimental studies. It is important to note that the numerical results are consistent with the experimental ones. Both of them indicate that a larger depth of flow corresponds to a lower surface temperature of a cable with a flaw.